AA2BBmap of a Equivariant Model

[zhanglw0521]

This has been more or less discussed and I am not even sure whether this is a good way to put it, but would it be useful to make the AA2BB map of an EQM a passable variable? By default it should be set to nothing (and thus the SVD procedure is performed) but when needed we can pass in those sparse coupling coefficients (e.g., obtained and saved in previous calculations or so) so that the EQM doesn't need to run the SVD redundantly.

[cortner]

I would make this an internal method but not a public one. I.e. it is something that we can use for efficiency when developing, but I wouldn't make it user-facing and thus requiring documentation and backward compat.